1. Field of the Invention
This invention relates to fuel control systems and, more particularly, to an improved method of estimating the air flow into an engine.
2. Background Art
An air-charge estimation algorithm is an important part of a spark-ignition engine management system. The estimate of the air flow into the engine is used to calculate the amount of fuel that needs to be injected so that the air-to-fuel ratio is kept close to the stoichiometric value for optimum Three Way Catalyst (TWC) performance.
In diesel engines, the air-to-fuel ratio must be maintained above a specified threshold to avoid the generation of visible smoke. At tip-ins, the EGR valve is typically closed and the control system calculates the amount of fuel that can be injected so that the air-to-fuel ratio stays at the threshold value. Inaccurate air-to-fuel ratio estimation in transients may result in either visible smoke emissions or detrimental consequences for torque response (increased turbo-lag).
A basic air-charge estimation algorithm relies on a speed-density equation that for a four cylinder engine has the form,             m      e        =                  η        v            ⁢                        n          e                2            ⁢              V        d            ⁢              p        RT              ,
where:
me is the mean-value of the flow into the engine, ne is the engine speed (in rps), xcex7v is the volumetric efficiency, xcfx81 is the intake manifold pressure, Vd is the total displaced cylinder volume, T is the intake manifold temperature, and R is the gas constant.
The volumetric efficiency map is typically calibrated on an engine dynamometer and stored in lookup tables as a function of engine operating conditions. In a conventional approach for a Variable Valve Timing (VVT) engine, xcex7v would be a function of valve timing, obtained as a result of elaborate calibration. The intake manifold pressure may be either measured by a pressure sensor (MAP) or, if there is no MAP sensor, estimated based on the intake manifold isothermic equation:             p      .        =                  RT                  V          IM                    ⁢              (                              m            th                    -                      m            e                          )              ,
where mth is the flow through the engine throttle (measured by a MAF sensor) and VIM is the intake manifold volume. This continuous time equation needs to be discretized for the implementation as follows:                                                         p              cal                        ⁡                          (                              k                +                1                            )                                =                                                    p                cal                            ⁡                              (                k                )                                      +                          Δ              ⁢                              xe2x80x83                            ⁢              T              ⁢                              RT                                  V                  IM                                            ⁢                              (                                                                            m                      th                                        ⁡                                          (                      k                      )                                                        -                                                            m                      e                                        ⁡                                          (                      k                      )                                                                      )                                                    ,                            (        1        )            
where xcex94T, is the sampling rate, mth(k) is the measured or estimated throttle flow and me(k) is the estimate of the flow into the engine based on the current measurement or estimate of the intake manifold pressure pcal(k). The variable pcal may be referred to as the modeled, estimated, or observed pressure. As is explained in more detail below, more elaborate schemes for air-charge estimation use the model in Equation (1) even if MAP sensor is available because useful information can be extracted from the error between the modeled pressure Pcal and the measured pressure p.
More elaborate schemes used in spark-ignition (SI) engines perform the following functions: compensate for the dynamic lag in the MAF sensor with a lead filter, see for example J. A. Cook, J. W. Grizzle, J. Sun, xe2x80x9cEngine Controlxe2x80x9d, in IEEE CONTROL HANDBOOK, CRC Press, Inc. 1996, pp 1261-1274; and J. W. Grizzle, J. Cook, W. Milam, xe2x80x9cImproved Cylinder Air Charge Estimation for Transient Air Fuel Ratio Controlxe2x80x9d, PROCEEDINGS OF 1994 AMERICAN CONTROL CONFERENCE, Baltimore, Md., June 1994, pp. 1568-1573; filter out the noise in the pressure and throttle flow measurements and adapt on-line the volumetric efficiency from the deviation between the actual pressure measurement and modeled pressure, see for example Y. W. Kim, G. Rizzoni, and V. Utkin, xe2x80x9cAutomotive Engine Diagnosis and Control via Nonlinear Estimationxe2x80x9d, IEEE CONTROL SYSTEMS MAGAZINE, October 1998, pp. 84-99; and T. C. Tseng, and W. K. Cheng, xe2x80x9cAn Adaptive Air-Fuel Ratio Controller for SI Engine Throttle Transientsxe2x80x9d, SAE PAPER 1999-01-0552. The adaptation is needed to compensate for engine aging as well as for other uncertainties (in transient operation). For engines without an electronic throttle, an estimate of the flow into the engine needs to be known several events in advance. In these cases, a predictive algorithm for the throttle position may be employed. See, for example, M. Jankovic, S. Magner, xe2x80x9cAir-Charge Estimation and Prediction in Spark Ignition Internal Combustion Enginesxe2x80x9d, PROCEEDINGS OF 1999 AMERICAN CONTROL CONFERENCE, San Diego, Calif.
In a typical embodiment of the schemes in the prior art, two low pass filters, on intake manifold pressure and throttle flow, may be employed to filter out the noise and periodic signal oscillation at the engine firing frequency. One dynamic filter would be used as a lead filter to speed up the dynamics of the MAF sensor. One dynamic filter would be used for the intake manifold pressure model and one integrator would be utilized to adjust the estimate of the volumetric efficiency as an integral of the error between the measured and estimated intake manifold pressure. This is a total of five filters.
It is an object of the present invention to provide an improved air-charge estimation algorithm.
It is another object of the present invention to provide an improved air-charge estimation algorithm that enables tighter air-to-fuel ratio control in SI engines.
It is a further object of the present invention is to provide an improved air-charge estimation algorithm that enables least turbo-lag to be achieved without generating visible smoke.
In accordance with the present invention, a method and system for estimating air flow into an engine is proposed that accomplishes the above steps of MAF sensor speedup, noise filtering and on-line volumetric efficiency estimation but uses only three dynamic filters. This reduces the implementation complexity of the air charge algorithm.
The mechanism for on-line volumetric efficiency estimation provided in the present invention is of differential type as opposed to the integral type algorithms employed in Kim and Tseng. The main advantage of the differential type algorithm of the present invention is that the correct estimate of the flow into the engine is provided even during fast changes in engine operation. In particular, in SI engines with VVT, valve timing changes would have a substantial influence on the air-charge. The proposed algorithm estimates the air-charge accurately even during fast VVT transitions, relying on no (or reduced amount of) information about VVT position or air-charge dependence on valve timing. Integral-type algorithms that adapt the volumetric efficiency are too slow to adjust to such rapid changes in the engine operation. Because no detailed information about the dependence of the air-charge on valve timing is required, the calibration complexity is reduced in the present invention.
More particularly, in accordance with the present invention, the flow into the engine is estimated via a speed-density calculation wherein the volumetric efficiency is estimated on-line. There are three interconnected observers in the estimation scheme. An observer is an algorithm for estimating the state of a parameter in a system from output measurements. The first observer estimates the flow through the throttle based on the signal from a mass air flow sensor (MAF). It essentially acts as a compensator for the MAF sensor dynamics. The second observer estimates the intake manifold pressure using the ideal gas law and the signal from an intake manifold absolute pressure (MAP) sensor. This second observer acts as a filter for the noise and periodic oscillations at engine firing frequency contained in the MAP sensor signal and the MAF signals. The third observer estimates the volumetric efficiency and provides an estimate of the air flow into the engine.